• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

Consolidated Bracketology Thread 3/12/23 updates

Decent slate of ACC games today for teams in the mid-to-upper tier of the league

Miami at NC State - Miami leads the league in 2-point %, blocks and steals. This will be a decent win for somebody
Duke at Clemson - Duke is winless away from home except for @BC. If Clemson is going to win the league, they need this one
Pitt at GT - can Pitt TCOB on the road in a mostly dead atmosphere? (RIP, ThrillerDome)
WF at BC - THE rivalry, 'nuff said
 
deacdixieboy said:
PAC-12 has two god awful teams with Oregon State and Cal.
Click to expand...

Mizzou coach Dennis Gates played at Cal and was a former assistant. He interviewed for the Cal job in 2019. They passed over him to hire retread Mark Fox. Gates took the Cleveland St job then got the Mizzou job. He’s 13-3 and ranked #20.
 
Wake goes from 81 to 75 in the NET.
Also 81 to 75 in Kenpom.

Apparently NET’s secret formula is has a lot of kenpom bias in it toward margin of victory.

Gotta wonder how much coach looks and those stats tonight and has to coach the team to them.

Meanwhile it looks like the old school RPI has us in the 20’s. Used to be the ultimate measure of success not so long ago.
 
Wakeforest22890 said:
Yes margin of victory matters. Beating a team by 20 instead of 2 is a big deal and should be
Click to expand...

For sure it matters

The should be is debatable but it is what it is.

Guys who are more on the fringe of the rotation are now completely out because you can’t afford to win by a small margin. As opposed to the old “just win baby” and maybe even develop some guys for later.

Again - I get it and it is what it is. It’s the new reality that we move forward with. But “should” is debatable. Perhaps a discussion for another thread.

Let’s keep moving up. Big opportunity this week.
 
To be fair the discussion has already happened multiple times a year on multiple threads. We know that how efficient you are on offense and defense per possession is the best predictive indicator over time of how good a team is. Over the course of one game winning margin is an excellent proxy for that.

If one team goes 30-0 and wins every game by 1 and another plays the same schedule, goes 30-0 and wins every game by 20 points we have a pretty good indicator that team B is better even if they’re equal record wise.

NET provides a bit of a boost in the ratings on the “resume” front by including something called TVI (which incorporates who won each game). Predictive metrics don’t really care who wins one off sample sizes, just how efficient you are on a schedule adjusted basis over the course of the season
 
Apologize if the discussion has been exhausted but it’s that time of year again for those newer to the discussion:

Two factors that I would add to your explanation that annoy me are:

(1) the NET is published on the NCAA website so that sort of endorsement means something, we can’t assume from that fact that it’s just one measure of many. They are telling us it’s more than that.
(2) the quads are arbitrary. Someone arbitrarily defined them. That’s a problem - to at least some degree.
 
lazydeac said:
Apologize if the discussion has been exhausted but it’s that time of year again for those newer to the discussion:

Two factors that I would add to your explanation that annoy me are:

(1) the NET is published on the NCAA website so that sort of endorsement means something, we can’t assume from that fact that it’s just one measure of many. They are telling us it’s more than that.
(2) the quads are arbitrary. Someone arbitrarily defined them. That’s a problem - to at least some degree.
Click to expand...

100% agree on point 2. As to 1, the NET is the NCAA’s system so that’s why it’s there. It’s one of several numbers that will appear on a team sheet but does have more importance since it’s the system they create the quadrants from
 
Sunday’s limited action results in:
1 in the Kenpom to 74
1 in the NET to 76

Michigan picks up a win over northwestern to pass us in NET.
 
Wakeforest22890 said:
If one team goes 30-0 and wins every game by 1 and another plays the same schedule, goes 30-0 and wins every game by 20 points we have a pretty good indicator that team B is better even if they’re equal record wise.
Click to expand...

And by this logic, if you have 4 teams and they all play each other, and the results are that teams A, B, and C all beat team D by 50 while the A/B/C games all came down to the the last shot but C lost both - why would you split the four into a "good" bucket of A/B and a "bad" bucket of C/D?

The problem is this half-in, half-out approach. Arbitrarily choose bucket sizes and home/away/neutral skews, only use metrics within that construct... It's fine at identifying the best and worst teams, does a garbage job trying to figure out the middle.
 
NET is one piece of the puzzle that the committee uses at the end of the day. It's an imperfect system, but does a pretty good job of getting most teams where they should be.

There's a reason Wake is between 65-75 in basically every predictive metric despite our 13-5 record.
 
DCDeac said:
And by this logic, if you have 4 teams and they all play each other, and the results are that teams A, B, and C all beat team D by 50 while the A/B/C games all came down to the the last shot but C lost both - why would you split the four into a "good" bucket of A/B and a "bad" bucket of C/D?

The problem is this half-in, half-out approach. Arbitrarily choose bucket sizes and home/away/neutral skews, only use metrics within that construct... It's fine at identifying the best and worst teams, does a garbage job trying to figure out the middle.
Click to expand...

I don’t understand the second paragraph here.

But from the results of the first paragraph, metrics would have A/B/C as essentially equal teams with D well behind them. I agree that the quadrant system from NET sucks and is a weird way of attempting to extrapolate data from the underlying numbers.

There should be more than 4 quadrants (octants maybe???). Not that Q4 breakdowns matter but that it counts equal for buckets when you beat the 201 team (basically BC) and last place - when BC would be favored by double digits on a neutral court - is super weird.
 
It's really, really dumb when you get a huge boost for beating the 30th team in the NET on your home court instead of the 31st team in NET on your home court.
 
Take a ridiculously simplified example.

If Team A and Team B play the same schedule, with the exact same number of possessions in each game, and

Team A wins every game by 1 (average efficiency +1)
Team B wins 20 games by 10, loses 10 games by 5 (average efficiency +5)

Then the predictive metric will like Team B by 4 points over Team A. That's all great...but you play the game to win the game, which Team A did a far better job of doing. Allegedly NET has some secret sauce to adjust for this. I don't think it should be a secret. I'd be interested to know which of those two teams NET prefers.
 
You must log in or register to reply here.
Back
Top